Women’s Voice, Agency, & Participation Research Series 2013 No.1 82534 Engaging men and boys in advancing women’s agency: Where we stand and new directions Paul J. Fleming Gary Barker Jennifer McCleary-Sills Matthew Morton This paper was commissioned by the World Bank Group to help inform a forthcoming report on women’s voice, agency, and participation. It does not necessarily reflect the views and research of the World Bank Group. Feedback and comments are welcome at: genderandagency@worldbank.org More details about the report are available at: www.worldbank.org/gender/agency 1 Table of Contents INTRODUCTION ............................................................................................................................................. 4 Gender norms and women’s agency ............................................................................................................. 5 What are gender norms and why do they matter? .................................................................................... 5 Intergenerational Transmission of Gender Norms ................................................................................... 9 Changeability of Gender Norms ............................................................................................................. 10 Men’s varied roles and women’s agency .................................................................................................... 10 Men and the five domains of agency ....................................................................................................... 11 Men as Perpetrators of Violence ............................................................................................................ 12 Men as Victims of Violence..................................................................................................................... 15 Men as advocates for preventing violence against women ..................................................................... 16 Examining male pay-offs to women’s agency and gender equality ............................................................ 17 Costs of gender inequality and violence against women for men ........................................................... 17 Benefits of gender equality for men ........................................................................................................ 18 Identifying new directions and strategies .................................................................................................... 18 METHODS...................................................................................................................................................... 19 Settings and Procedures .............................................................................................................................. 19 Measures ..................................................................................................................................................... 22 Demographic variables .......................................................................................................................... 22 Outcome Variables ................................................................................................................................. 22 Attitudinal and Behavioral Variables ..................................................................................................... 23 Data Analysis .............................................................................................................................................. 24 Limitations .................................................................................................................................................. 25 RESULTS........................................................................................................................................................ 26 Attitudes towards gender equality ............................................................................................................... 28 Attitudes and practices regarding women’s roles and responsibilities........................................................ 29 Men’s attitudes regarding violence against women .................................................................................... 31 Perpetration of Violence against women .................................................................................................... 35 Men reporting they would prevent violence against women ...................................................................... 43 DISCUSSION ................................................................................................................................................. 47 Key Findings ............................................................................................................................................... 48 Support from men for gender equality and women’s agency ................................................................. 48 The role of education and other factors.................................................................................................. 48 2 Violence against women still common but can be prevented ................................................................. 49 Risk factors for perpetrating violence .................................................................................................... 49 Men can be advocates against violence .................................................................................................. 51 Comparison with findings from Peers for Prevention (P4P)....................................................................... 51 Promising Directions for Interventions and Policy ..................................................................................... 54 Community-level interventions ............................................................................................................... 54 Mass media ............................................................................................................................................. 56 Leveraging Policy ................................................................................................................................... 56 Future Research Directions ......................................................................................................................... 60 CONCLUSIONS ............................................................................................................................................. 61 REFERENCES ................................................................................................................................................ 63 Tables and Figures Table 2 .............................................................................................................................................................................................. 28 Table 3 .............................................................................................................................................................................................. 30 Table 4 .............................................................................................................................................................................................. 30 Table 5 .............................................................................................................................................................................................. 31 Table 6 .............................................................................................................................................................................................. 33 Table 7 .............................................................................................................................................................................................. 33 Table 8 .............................................................................................................................................................................................. 34 Table 10 ............................................................................................................................................................................................ 40 Table 11 ............................................................................................................................................................................................ 42 Table 12 ............................................................................................................................................................................................ 46 Table 13 ............................................................................................................................................................................................ 46 Table 14 ............................................................................................................................................................................................ 47 Table 15 ............................................................................................................................................................................................ 54 Table 16 ............................................................................................................................................................................................ 82 Table 17 ............................................................................................................................................................................................ 83 Figure 1 ............................................................................................................................................................................................. 21 Figure 2 ............................................................................................................................................................................................. 33 Figure 3 ............................................................................................................................................................................................. 34 Figure 4 ............................................................................................................................................................................................. 37 Figure 5 ............................................................................................................................................................................................. 38 Figure 6 ............................................................................................................................................................................................. 41 Figure 7 ............................................................................................................................................................................................. 43 Figure 8 ............................................................................................................................................................................................. 58 3 ABSTRACT Despite advances in gender equality, women and girls still face disadvantages and limits on their agency. Men and women are both subject to gender norms that influence their behaviors and these norms can constrain women’s agency and can encourage men to adopt behaviors, including sometimes violent behaviors, which further constrain women’s agency. Men and boys can be key stakeholders and allies to increase women’s agency and this paper focuses on examining men’s attitudes and behaviors related to gender equality and violence perpetration to better understand how to engage men and boys as. To do so, we use data that were collected from men and women from eight countries (Bosnia, Brazil, Chile, Croatia, Democratic Republic of Congo, India, Mexico, and Rwanda) as part of the International Men and Gender Equality Survey (IMAGES). We found that there is wide variation across countries in men's support for gender equality, equal roles for men and women, and acceptability of violence against women. In multivariate analysis examining predictors of men ever perpetrating physical violence against a partner, we found that older age, witnessing partner violence against one’s mother, permissive attitudes towards violence against women, having inequitable attitudes, and having been involved in fights generally were all associated with a higher likelihood of perpetrating violence. A majority of men is willing to intervene if they witness violence against a woman, and men who do not support violence against women, are not violent generally, and are aware of laws prohibiting violence against women are more likely to intervene. We interpret these findings and identify key knowledge gaps and directions for future research, public policies, evaluation, and programming Key findings:  In most countries, male perpetrators of violence are more likely to be depressed or engage in binge drinking than non-perpetrators.  Witnessing one’s mother being abused by a partner is one of the strongest predictors of ever perpetrating violence, suggesting that efforts should focus on breaking the intergenerational transmission of norms and violence.  Being involved with violent fights generally is a significant predictor of ever perpetrating violence, suggesting that programs and policies reducing violence generally may also have an effect on violence specifically against women.  A majority of men is willing to intervene upon witnessing violence against a woman, and men who do not support violence against women, are not violent generally, and who are aware of laws prohibiting violence against women are more likely to intervene. 4 INTRODUCTION The World Development Report 2012 (WDR 2012) highlights the important economic, health, and social gains that can be achieved by increasing gender equality worldwide (World Bank 2011b). Despite advances in gender equality, women and girls still face disadvantages and limits on their ability to make choices for themselves. The WDR 2012 identifies increasing women’s agency, or the “ability to make choices to achieve desired outcomes” (p. 3), as one of four priority area for achieving gender equality. And while the report focuses on women and girls, increasing their agency cannot occur in isolation. Men and women both are subject to gender norms that influence their behaviors. These norms can constrain women’s agency and can encourage men to adopt behaviors, including sometimes violent behaviors, which further constrain women’s agency. The constraints associated with gender norms typically lead to sub-optimal health, livelihood, and well-being outcomes for men, women, boys, and girls. This paper focuses on examining men’s attitudes and behaviors related to gender equality and violence perpetration to better understand how to engage men and boys as key stakeholders and allies to increase women’s agency. This background paper aims to review and deepen the evidence base on the role of men and boys in advancing women’s agency. The objectives of this paper include the following:  Review background literature on the influence of gender norms on men and women’s behaviors and attitudes, discussing the role of men and boys in each of the domains of women’s agency identified in the WDR 2012.  Review men’s use of violence against women and examine the benefits of gender equality and women’s agency for men.  Present new analyses from the International Men and Gender Equality Survey (IMAGES) that focus on men’s attitudes and practices, and their roles as perpetrators of violence and as allies in promoting women’s agency in eight countries.  Outline key knowledge gaps and directions for future research, public policies, evaluation, and programming. Gender norms and women’s agency What are gender norms and why do they matter? Gender norms are broadly understood by members of a population and are considered “those qualities of femaleness and maleness that develop as a result of socialization rather than biological 5 predisposition”(Boles & Hoeveler 2004). Distinct norms of behaviors for men and women are socially constructed, which means that gender is created through patterns of social interactions, not determined biologically (Connell 1987). These norms of behavior for men and women are particularly powerful because deviations can be punished through social exclusion, ostracism, or sometimes violence (Dorais & Lajeunesse 2004). For a woman who breaks restrictive norms of femininity, for instance, social or physical punishment can include ridicule by peers, expulsion from her home, or violence (Macmillan & Gartner 1999; Reidy, Shirk, Sloan, & Zeichner 2009). For example, one Tanzania woman participating in a focus group on violence against women described common punishment for not obeying her husband: “It is very common if you refuse his orders you will be beaten, when he denies to start a business and you did it anyway, you will be beaten.” (McCleary-Sills et al. 2013) These consequences serve to limit women’s agency. In the most gender unequal societies, the negative consequences are so great that most women have little room to deviate from the societal norms and make choices for themselves without serious repercussions. Men’s actions and behaviors are subject to norms of masculinity in the same way that women are subject to norms of femininity. Examinations of gender norms often focus on the limitations placed on women’s agency, but men also are limited to the behaviors and practices that are deemed socially acceptable (Connell, 1995b). Examinations of men’s gender norms have looked at a society’s constructed “hegemonic masculinity,” defined by R.W. Connell’s seminal book Masculinities, as the form of masculinity that is recognized as the most dominant in a society’s pattern of gender relations (Connell 1995b). This hegemonic form of masculinity is often characterized by being aggressive, risk-taking, virile, unemotional, and dominant over women. Men’s social position depends in part on their ability to outwardly conform to the standards of this type of masculinity. But this “hegemonic” form of masculinity is an idealized version that few men can realistically achieve. As Connell writes: “The number of men rigorously practicing the hegemonic pattern in its entirety may be quite small. Yet the majority of men gain from its hegemony, since they benefit from the patriarchal dividend, the advantage men in general gain from the overall subordination of women.” Hegemonic masculinity therefore permeates throughout the majority of males in the society even though individual males may not be performing the masculine ideal. Importantly, men complicit in the practice of hegemonic masculinity do not necessarily actively support the subordination of women. However, the entire patriarchal social and power structure gives men power and status over 6 women, so most men are incentivized to not fight against it (Connell 1995b). The influence this system of power has on almost all males in a society is extremely important to the resulting behaviors of men (Courtenay 2000). As men weigh their decisions (consciously or subconsciously), their position in this power structure, and their desire to maintain position or advance, will typically play a role in how they behave in social situations (Courtenay 2000). The social consequences for men not adhering to societal gender norms are increasingly recognized as influential in men’s behaviors (Hyde, Drennan, Howlett, & Brady 2009; Levant, Wilmer, Williams, Smalley, & Noronha 2009). While men’s greater decision-making power does typically give men more agency than women, their decisions are constrained by pressures to be considered masculine. In some settings, men’s decision to use contraceptives is constrained by a need to conform to local norms of masculinity. One Tanzanian woman interviewed for the WDR 2012 qualitative study reported, “You cannot tell men to use birth control; they want children. The more they have, the more manly they appear to be” (Munoz Boudet, Petesch, Turk, & Thumala 2012) p. 90). Likewise, in certain cultures, a man’s decision to take on a prominent care-giving role with his family’s children may result in a loss of social status (Richter & Morrell 2006; van den Berg et al. 2013). Although men may have greater decision-making power than women, their decisions are still constrained by the negative social consequences of certain decisions that deviate from the hegemonic masculinity. In this way, men’s decisions are linked to their projection of a masculine identity for their community and peers (Connell 1995b; Courtenay 2000; S. L. Dworkin, R. E. Fullilove, & D. Peacock 2009). One of the most common roles for men across cultures is to be the provider and protector of their family (Connell 1995b; Gilmore 1990). Thus, men who are able to support their families are fulfilling a primary cultural duty for men and projecting their masculinity for their community. Men who are unable to provide for their family may find alternative methods to demonstrate their masculinity (Barker 2005). These men sometimes use their behaviors, such as sexual activity, capacity for drinking, or shows of force, to demonstrate their masculinity for their peers (Courtenay 2000). Men are often culturally obligated to project a masculine image since the consequences for individual men who are perceived as non-masculine or feminine can be great, from social ostracism (Cohan 2009) to death by violence (Dorais & Lajeunesse 2004; Kimmel & Mahler 2003). Notably, although men as a group typically hold greater power than women, this does not imply that all men are powerful. In fact, poor men and minority men are often marginalized and left out of 7 the traditional power structures (Courtenay 2000; D. R. Williams 2003). They may still have more power or authority than the women of their lives, but their power in society is limited. For example, many men across settings in the WDR 2012 qualitative study reported having insufficient power and freedom to make strategic decisions that could improve their lives and the lives of their family(Munoz Boudet et al. 2012). Men can sometimes find this perceived lack of power frustrating and adopt certain behaviors (e.g. violence, sexual behaviors) that gives them a sense of power over others (Barker 2005). The social dimensions of masculine behavior play out in households across the globe. In many settings, men’s higher status affords them greater decision-making power than women (Connell 1987, 1995a; Wingood & DiClemente 2000). Men’s household decision-making power influences an array of health and well-being issues affecting men, women, and children, including sexual health (Campbell 1995), nutrition (Kennedy & Peters 1992), mental health (L. Heise, Ellsberg, & Gottmoeller 2002), economic well-being (Okojie 1994), and health care utilization (Okojie 1994). Men often have economic control over the provision of health resources for the family (Doyal 2000) and can sometimes have more decision-making power than women regarding the use of contraceptive methods (Cabral, Pulley, Artz, Brill, & Macaluso 1998). Men’s behaviors can influence their own health status (Pinkhasov et al. 2010). For example, because men are socialized to project an invulnerable image, men are less likely to utilize healthcare services that might reveal their vulnerabilities (Courtenay 2000). Norms of masculinity encourage aggression and subordination of women, both factors that can results in violence against partners or children. Men’s perpetration of violence against women is enabled by norms of masculinity and prevailing norms of gender equality in most societies (L. L. Heise 1998). A review of research on the role of masculinity in partner violence presented evidence on different domains of masculinity and male gender norms that influence perpetration of violence (Moore & Stuart, 2005). The review showed that various research studies have demonstrated that men who hold more traditional gender role ideologies (i.e. distinct roles for men and women) are more likely to perpetrate violence (R. Jewkes, Sikweyiya, Morrell, & Dunkle 2011; Levtov, Barker, Contreras, Heilman, & Verma, Forthcoming; Santana, Raj, Decker, La Marche, & Silverman 2006). Additionally, men’s gender role strain has been identified as a risk factor for perpetrating violence (Copenhaver, Lash, & Eisler, 2000; Franchina, Eisler, & Moore, 2001; Jakupcak, Lisak, & Roemer 2002). Gender role strain refers to men’s feelings about their ability to conform to normative ideas about what it means to be a man (Pleck 1995). Societies with greater gender inequities are more likely to teach young men a 8 traditional gender role ideology and increase pressure that men act in traditionally masculine ways. Both of these factors, the research shows, increase the likelihood that a man will perpetrate violence against an intimate partner. Intergenerational transmission of gender norms Gender norms are transmitted from generation to generation (Farré & Vella 2007). Gender norms are replicated by social observation of behaviors, particularly as children and youth observe their elders. As such, norms influence behaviors, and behaviors influence norms. For example, as more men take on a care-giving role, the gender norms for men will likely start to shift slightly to include care-giving. And, as care-giving becomes a norm for men, more men will begin to practice this behavior and the cycle will continue. This shift in behaviors and norms is in part driven by psychosocial concepts from the Social Learning Theory and its subsequent version, Social Cognitive Theory (Bandura 2001). These behavioral theories have established that individuals learn how to behave socially through observing and imitating important others in their social environment. This observation and imitation occurs throughout the lifespan, but can be particular important for children and youth. Often, children learn from their parents what are appropriate behaviors for boys and girls, men and women, and these lessons can impact their behaviors and attitudes throughout their lifetime. Throughout childhood, children typically receive positive consequences for conforming to the way that their parents believe a boy or girl should act, and negative consequences for any deviation. Farre and Vella found that mothers with positive views about women’s participation in the labor force were more likely to have children who also viewed women’s labor participation positively when they reached adulthood (Farré & Vella 2007). Further, they found that both their daughters and their son’s wives were more likely to participate in the labor force if the mothers had positive views towards women’s participation (Farré & Vella 2007). Previous analysis of IMAGES data shows that, in most countries, men whose fathers participated equally in domestic duties were significantly more likely to participate in domestic duties than men whose fathers did not participate equally (Barker et al. 2011). This observational learning also applies for other adults in the child’s life, such as teachers, relatives, or other respected adults. Experiences in childhood can have a lasting impact on that child’s path into adulthood. Harmful norms, such as those encouraging violence, can also be transmitted across generations. A meta-analytic review of 39 published research studies on the intergenerational transmission of 9 partner violence demonstrated that children who witness intra-parental violence are themselves more likely to be involved in violent relationships in adulthood (Stith et al. 2000).These studies all point to the importance of childhood experience on later attitudes towards gender equality and perpetration of violence. Changeability of gender norms Despite the ease by which gender norms are transmitted and passed from generation to generation, gender norms can and do change. As mentioned before, norms are derived from patterns of behaviors and slow shifts in behaviors can produce new gender norms. The movement for women’s rights and equality has been ongoing for over a century, during which standards and norms for women have changed drastically in many parts of the world. While a similar radical transformation of gender norms has not yet occurred for men, there is evidence to show that men’s attitudes and practices have changed from previous generations. For example, IMAGES analysis has shown that younger generations of men are more supportive of gender equality and more likely to engage in household tasks (Barker et al. 2011). Additionally, many social programs have aimed to increase gender equality by changing or challenging some of the gender norms that facilitate inequalities. Many international organizations are focusing programs on achieving greater gender equality, including programs promoting the education of girls and microfinance programs that encourage women’s economic independence (Pronyk et al. 2008). From another angle, Instituto Promundo, a Brazilian-based international NGO, has created, evaluated and disseminated programs aimed at helping young men question masculine norms that promote violence and objectification of women (Pulerwitz, Michaelis, Verma, & Weiss, 2010). When men are asked to critically think about and challenge the assumptions of a gender unequal society, this program has found that men can become less complicit in the existing unequal power structure. Other programs using these strategies include the International Center for Research on Women’s Parivartan (Das, Ghosh, Miller, O'Connor, & Verma 2012), a school-based intervention in India where trained coaches promoted violence prevention, and SASA! (Abramsky et al. 2012), a cluster-randomized control trial in Uganda that uses community mobilization to transform community gender norms around violence. These programs, as well as grassroots movements and policy changes, help facilitate a shift towards greater gender equality. Men’s varied roles and women’s agency Gender norms impose limits to agency by delimiting appropriate behaviors for men and women. Women’s agency is defined in the 2012 World Development Report as the “ability to make 10 effective choices and to transform those choices into desired outcomes.” (World Bank 2011b p. 150). The social structure of our global societies typically places more limits on women’s agency than on men’s. The WDR identified five main outcomes, or “expressions,” that characterize women’s agency: 1) control over resources, 2) ability to move freely, 3) decision making over family formation, 4) freedom from risk of violence, and 5) ability to have a voice in society and influence policy. In this section, we discuss the ways in which men play a role in limiting and enabling women’s agency. We start by discussing the various ways by which men can limit a woman’s agency, then examine men’s relationship with violence both as perpetrators and as victims, and finally demonstrate how men also enable women’s agency and can play an active role in reducing gender inequalities. Men and the five expressions of agency Men can play a role in limiting women’s agency for each of the five expressions at higher levels of the socio-ecological framework1 (Bronfenbrenner 1979; L. L. Heise, 1998). At the household level, the men in a household can play an important role to enforce limits on women’s agency. Enforcement of these limits does not necessarily imply physical force but can include a spoken or unspoken threat of social exclusion or removal from the household for deviating from the norm. These types of violence are often considered emotional, psychological, or economic abuse (Garcia- Moreno, Jansen, Ellsberg, Heise, & Watts 2005). A husband may use economic violence to limit a wife’s “control over resources” by limiting her access to family income, restricting her decision- making power for purchases, or limiting access to legal documentation or identification (Raj, Silverman, McCleary-Sills, & Liu 2005). For example, a qualitative study conducted in 19 countries for the WDR 2012 found that over one fifth of participants reported the husband controlling the wife’s earnings (World Bank 2011a). Male household members may restrict female member’s “ability to move freely” by disallowing them to work outside the home or congregate in spaces where other men are present. If men do not enforce these rules through physical violence, they may use emotional violence such as humiliation or threats to impose their rules (Garcia- Moreno et al. 2005). Men can limit women's agency at other levels of the socio-ecological framework as well. For example, at the organizational/institutional level, men may control institutions, such as schools, health centers, and law enforcement, which impose rules and regulations (both formal policies and 1 The socio-ecological framework identifies multiple levels of influence on individuals, including interpersonal, organizational/institutional, community, and policy levels. It has been applied broadly to a variety of issues, including analysis of gender inequalities. 11 informal norms/rules) that further limit a woman’s ability to exercise agency. At the community level, gender norms around gendered public and private spaces that are enforced by men (and women) can limit a woman’s ability to enter spaces where decisions are made. And finally, at the policy level, most policymakers and enforcers across the globe are men and are less likely to establish laws that are equally fair to men and women (UN 2010). A variety of public policies related to domestic violence, childcare or employment serve to limit women’s agency by not protecting women’s rights and opportunities (Kiluva-Ndunda 2001). Husbands and fathers can also restrict a woman’s “ability to make decisions about family formation” (Dodoo 1998; Dodoo & Frost 2008). For example, a husband may limit a wife’s access to contraceptive methods or information about those methods. Additionally, in more egregious cases men may thwart a woman’s attempt to avoid pregnancy, including intentionally sabotaging a contraceptive method, refusing to use or allow the use of contraception, and exerting control through threats, accusations of infidelity, or forcing her to have sex when she is unwilling to do so (Clark et al. 2008; Miller et al. 2010). In the 19 countries researched for the WDR 2012 report, between 20 percent and 40 percent of young men and women interviewed reported contraceptive use being the man’s decision. Additionally, a father may make decisions for his daughter on who she will marry or when she will have children (UNFPA 2012).Women and girls in a household may also have their agency limited by not being “free from violence.” They may be subjected to physical, emotional, or sexual violence carried out by the men or boys in their household. In a multi-country study carried out by the WHO, between 20 percent and 75 percent of women reported experiencing emotional abuse, 1-21 percent of women reported experiencing child sexual abuse, 13-61 percent reported being physically abused by a partner, and between 6 percent and 59 percent reported being a victim of sexual violence by a partner (Garcia-Moreno et al. 2005). And, finally, men may restrict a woman’s “ability to have a voice in society” by denying her permission to participate or educational opportunities that would allow her to participate fully in community or political activities. Put simply, men have many tools of socially sanctioned cooptation and coercion at their disposal with which to constrain women’s agency. Men as perpetrators of violence While one of the expressions of women’s agency is “freedom from violence,” violence is also an underlying threat that that enforces each of the other expressions. Violence can be perpetrated through emotional, physical, sexual, or economic violence (Garcia-Moreno et al. 2005). Men who perpetrate violence against women may be doing so as a means to restrict women’s agency in one 12 of the five domains, or as a means to assert his dominance and masculinity. In any case, each of the types of violence is considered to stem from men’s attempt to control and have power over women. Women’s agency is restricted when men employ violence, including emotional violence, to intimidate and scare women from asserting or enacting their agency. Between 15 percent and 71 percent of women in the World Health Organization’s multi-country study report experiencing physical or sexual violence by an intimate partner at some point in their lives (most sites were between 30 percent and 60 percent) (Garcia-Moreno et al. 2005). In previous IMAGES survey analysis of 6 countries, male-reported perpetration of physical violence against a partner was between 17 percent in Mexico and 39 percent in Rwanda (Barker et al. 2011). Though more than 75 percent of violence against women reported (by women) globally is that committed by male intimate partners, non-partner physical and sexual violence is also a problem. Perpetrators of non-partner physical violence include fathers, other family members (male and female), and teachers. Estimates of the prevalence of women 15 years or older who have experienced non- partner physical violence ranges from less than 10 percent to 62 percent (Garcia-Moreno et al. 2005). Female-reported perpetrators of sexual violence include largely boyfriends and partners. In a multi-country study, only 1 percent to 12 percent of sexual violence reported by women was perpetrated by a stranger to the victim (Garcia-Moreno et al. 2005). Men’s perpetration of violence against women (VAW) 2 results from a complex, interconnected ecology of psychological, economic, and sociological factors (L. L. Heise 1998). There are societal level factors, such as gender inequalities and patriarchal family structures, that facilitate a social environment that enables violence against women. But, not all men within gender unequal societies perpetrate violence, and thus individual risk-factors also play a role in men’s perpetration of VAW. A meta-analysis conducted on men’s perpetration of violence against a female intimate partner (married or cohabiting partner) identifies certain characteristics that are correlated with men’s perpetration of physical violence against their partner (Stith, Smith, Penn, Ward, & Tritt 2004). The factors most associated with perpetration were low marital satisfaction, illicit drug use, and attitudes condoning marital violence. Other important factors included traditional sex-role ideology, alcohol 2 Violence against women (VAW), as defined by The United Nations Declaration on the Elimination of Violence against Women, is “Any act of gender-based violence that results in, or is likely to result in, physical, sexual or psychological harm or suffering to women, including threats of such acts, coercion or arbitrary deprivations of liberty, whether occurring in public or in private life” (UN 1993). 13 use, depression, and career/life stress. Two separate meta-analyses identify witnessing abuse as a child as a moderate predictor of abuse perpetration in adulthood (Gil-Gonzalez, Vives-Cases, Ruiz, Carrasco-Portino, & Alvarez-Dardet 2008; Stith et al. 2000). Preliminary evidence from the IMAGES study indicates that violence perpetration is associated with greater support for inequitable gender attitudes (Levtov et al., forthcoming). Certain contexts facilitate perpetration of VAW, such as setting with unenforced or limited laws preventing VAW (L. L. Heise, Raikes, Watts, & Zwi 1994). Areas of conflict or post-conflict typically have much higher rates of violence against women, especially sexual violence. These higher rates are partially due to the increased impunity of perpetrators as social institutions that prevent VAW break down or become ineffective. This can continue to cause increased rates of violence in post-conflict settings if courts and institutions responsible for preventing violence are not established or repaired (UN Women 2013). In addition to using physical violence, men, and sometimes other women, can enforce limits to women’s autonomy through threats and emotional violence that humiliate, ostracize, or otherwise diminish the woman’s social status (Tjaden & Thoennes 2000). This emotional violence is often perpetrated by male partners as a means to control or dominate women (Tjaden & Thoennes 2000). Even if some men do not perpetrate these forms of physical and emotional violence against women, all women in a community are affected by the underlying possibility of violence and threat to their emotional and physical well-being(Connell 1995a). While violence against women may only be performed by a minority of men (Barker et al. 2011), the presence of violence against women in a society provides an implicit enforcement of traditional social norms that limit women’s agency. In most countries, the majority of men are not violent against women but play a role in enabling violence. Non-violent men may not desire to limit the agency of the women in their lives, but many of these men do not fight against the higher level systems (i.e., institutions/organizations, and public policies) that are limiting the agency of the women in their families and community. Prominent gender scholar R.W. Connell describes this as a complicit version of masculinity where men are not in favor of fighting for male dominance over women, but they are also hesitant to fight against it. Connell describes most men’s complicity as being motivated by the patriarchal dividend, or “the advantage men in general gain from the overall subordination of women ” (Connell 1995b), p. 79). Connell continues: 14 “Marriage, fatherhood, and community life often involve extensive compromises with women rather than naked domination or an uncontested display of authority. A great many men who draw the patriarchal dividend also respect their wives and mothers, are never violent towards women, do their accustomed share of the housework.” Most men may compromise with and respect the women in their lives, but do not challenge the broader societal power structure that favors men. Thus, the limits on all women’s expression of agency stems from both a minority of men who perpetrate violence and a majority of men who are complicit with the current gender order (Connell 1987, 1995b). Of course, there are also men who actively contest VAW and limits on women’s agency (UN 2008). Men can challenge VAW in the course of their daily life, through intervening with a violent neighbor, for example, or through advocacy efforts that challenge gender inequalities. Men as victims of violence Some of men’s violence perpetration may be due to previous experiences of violence or witnessing violence. Boys can be victims of violence at the hands of their fathers, mothers, siblings, peers, or other adults in their lives. The IMAGES data from six countries found that many men reported being victims of violence in childhood:  Between 20 percent and 85 percent report having experienced psychological violence before the age of 18  between 26 percent and 67 percent report having experienced physical violence before 18  between 1 percent and 21 percent of men report having experienced sexual violence before 18 (Contreras et al. 2012) These childhood experiences can teach boys two important lessons that can have an effect on future behaviors. First, it models to boys that violence is a way to resolve frustrations, stress, or conflict. Second, when parents or older siblings are violent, it can teach a boy that it is acceptable to assert power through shows of force and violence (Contreras et al. 2012). Both of these factors can have detrimental effects as boys turn into men and are in relationships with women. In fact, men who experienced violence as children were also more likely to hold gender inequitable attitudes in adulthood (Contreras et al. 2012). Youth is also a particularly vulnerable time for boys as they make the transition from childhood to adulthood. Many male adolescents are subjected to a culture of violence among their male peers as young men try to prove their masculinity through shows of force (Barker 2005; Kimmel & Mahler 2003). Social status is particularly salient for youth and young men are often encouraged to defend their status/honor through violent fights. Boys and young men who are gay or do not conform to norms of masculinity can also be subjected to emotional and physical violence by their peers 15 (Dorais & Lajeunesse 2004; Kimmel & Mahler 2003) In some areas, young men are pressured to join gangs as a way of social belonging or protection (Barker 2005). These processes can lead to a cycle of victimization and perpetration where young men are victimized and encouraged to retaliate with further violence. Again, this formative experience of being encouraged to respond with violence to challenges to their manhood can lead to greater likelihood to use violence against women (Contreras et al. 2012). Additionally, men can be victims of violence from their intimate partners. Some statistics show that men are just as likely to be victims of intimate partner physical violence as women (Black et al. 2011; Swart, Seedat, Stevens, & Ricardo 2002). But men are much less likely to be physically harmed by violence perpetrated by women and less likely to report fearing their partner (WHO 2010; J. R. Williams, Ghandour, & Kub 2008). Perpetration by women against males may also be the result of women defending themselves against men's violence (J. R. Williams et al. 2008). Men as advocates for preventing violence against women Men can, and sometimes do, play a crucial role in fighting for women’s rights and protections against violence. In many settings most men are opposed to violence against women and are supportive of gender equality (Barker et al. 2011). Like limits on women’s agency, this can occur at multiple levels, from the household to the policy level. The first and most important step that men take to prevent violence against women is to not perpetrate any type of violence against the women and girls in their lives. But men also go beyond this to prevent other cases of male perpetrated violence against women. Fathers can raise their sons to understand the harm of perpetrating violence against women (van den Berg et al. 2013). Men can also intervene with other men who are perpetrating violence (Fabiano, Perkins, Berkowitz, Linkenbach, & Stark 2003). This could involve calling the police, breaking up the violence, or mobilizing neighbors to confront the man. Men also get involved in advocacy efforts that raise awareness about the harms of violence against women and promote gender equality (S. Dworkin, Hatcher, Colvin, & Peacock 2013; van den Berg et al. 2013). Men have organized marches, lobbying efforts, and community events to speak out against violence against women. For example, men in Burundi have begun an awareness campaign where men who have recognized the harmful effects of violence against women travel to other nearby villages to share their experiences of change from violence to non-violence (Wallacher 16 2012). In Fiji and Vanuatu, women’s organizations have trained Male Advocates who are local leaders who work with their communities to reduce family and sexual violence (Ellsberg, Heilman, Namy, Contreras, & Hayes 2012). These efforts can raise awareness and encourage policymakers to devote more resources to be dedicated to the prevention of violence. Some men who are in a position of power also have the capability to change laws and policies related to preventing violence against women. Male government officials have fought to increase women’s abilities to press charges against violent men, protections against sexual assaults, and programs to prevent men’s perpetration. Additionally, male leaders in a society can support the enforcement of laws preventing violence against women (UN 2008). Often, these actions by men are carried out in conjunction with women’s groups and can help contribute to societal change. Examining male pay-offs to women’s agency and gender equality Increasing women’s agency and empowering women does not need to result in a loss of agency and empowerment for men. Research demonstrates that there are costs for men associated with gender inequality and violence perpetration. Additionally, men often benefit from greater gender equality. Societies as a whole, including the men, often benefit from greater economic and political participation by women. Costs of gender inequality and violence against women for men At the individual level, men who perpetrate violence against women are more likely to suffer a variety of mental and physical health ailments (Barker et al. 2011). In Vietnam, both men and women in a partnership missed days of work when a man abused his partner (N. Duvvury & Carney 2012). While gender inequalities that give men decision-making power seem beneficial to men, they restrict men’s agency by creating strict roles for men and women. In places where gender inequalities are prevalent, men may be resistant of care-giving professions even though they could receive higher pay (C. L. Williams 1993). Further, inequalities establish different norms for men and women, which has resulted in higher burden of disease and higher rates of behavioral risk factors for diseases (Wang et al. 2012). Some of these risk factors include men’s higher rates of smoking tobacco, drinking alcohol, and lower rates of health-seeking behaviors, each of which is derived from gender norms associated with gender inequality (Hawkes & Buse 2013). Intimate partner violence has costs for the household that in turn affect the men living within them. One report conducted by the International Center for Research on Women (ICRW) identified the 17 costs to household of intimate partner violence (N. Duvvury, Grown, & Redner 2004). They noted that women who are victims of violence have lost wages, lost productivity, and their children may miss school more, all of which have negative impacts on the household (N. Duvvury et al. 2004). Benefits of gender equality for men Increasing women’s agency does not need to decrease men’s agency and can actually contribute to increased economic and psychological well-being of household members. The IMAGES data has shown that men who are in more gender equitable relationships are more likely to be happy and satisfied in their relationships (Barker et al. 2011), and that women who report that their male partners are more involved in caregiving and support equitable decision-making in the household are happier with their partners, including sexually happier. In fact, in most countries, men reported greater sexual satisfaction if they had gender equitable beliefs and if they reported having open communication with their partner. Additionally, men who participate actively as fathers are more likely to have better physical and mental health (Dykstra & Keizer 2009). In a multi-country qualitative study, men who had taken on greater caregiving roles reported benefits to their friendships, relationship with their children, and improved relationship to their spouse (Barker et al. 2012) The household earning potential is also increased when women are allowed to work and have the opportunity to earn an income. This can increase the opportunities for all members of the household, including men. Breaking down the inequalities and increasing women’s agency can improve the lives of men and women. Identifying new directions and strategies The evidence is clear that there are important benefits to men, women, and children associated with increased gender equality and increasing women’s agency. Further, throughout the world, there has been a gradual shift in men’s attitudes towards the role of women and gender equality. To identify key strategies to help facilitate this shift in gender norms for men and women and decrease violence against women, it is critical to better understand men’s current attitudes and practices. Given that research on women’s rights and empowerment often focuses on women, there is a significant gap in the literature around men’s attitudes and practices related to increasing women’s agency and reducing violence. In this paper, we analyze data from 8 low- and middle-income countries to answer the following research questions:  What are men’s attitudes towards gender equality, division of resources, and violence against women?  What are the key differences between men who perpetrate physical violence against women and men who do not? 18  What factors are associated with men reporting that they would actively prevent violence against women in their communities? METHODS The data used for these analyses come from the International Men and Gender Equality Survey (IMAGES) as part of the Men and Gender Equality Policy Project co-coordinated by the International Center for Research on Women (ICRW) and Instituto Promundo. The purpose of this ongoing data collection effort is to better understand men’s attitudes and practices related to gender equality. Most international survey research on women’s rights and gender equity focuses on women, or has limited data on men’s attitudes on the role and agency of women. The IMAGES project aimed to fill this critical gap by carrying out representative household surveys in a diverse set of countries. IMAGES is innovative in its focus on men. The only other study of men in multiple countries that focuses on gender equality and violence is the UN Multi-country Study on Men and Violence in Asia and the Pacific. The UN Multi-country Study was conducted in 6 different Asian countries and was informed by the IMAGES data collection. Both IMAGES and the UN Multi-country Study use similar sampling methodology (stratified random sampling of cities/villages). For the current study, we rely on the IMAGES study and compare a some of our findings to the UN Multi-country Study data in the discussion section. Settings and procedures The IMAGES surveys have been carried out in eight countries to date: Bosnia, Brazil, Chile, Croatia, Democratic Republic of Congo (DRC), India, Mexico, and Rwanda. (Note: data collection has also been completed in Mali and Malawi, but these data are not included here because they must still be processed and cleaned). IMAGES utilized a stratified random sample to select households (See Figure 1). Data from Bosnia and Rwanda are nationally representative, and the data from the rest of the countries are representative of the regions/cities where they were conducted. Researchers in Chile sampled from three metropolitan areas, Croatia from one metropolitan and two rural areas, India from two metropolitan areas, and Mexico from three metropolitan areas. The DRC survey intended to understand men’s attitudes and practices in a post - conflict setting and therefore sampled from an internally displaced persons camp and military base, and two nearby rural villages. As a result, the DRC sample is unique and should not be regarded as representative of the country as a whole. In each of the settings, men aged 18-59 were randomly selected from sampled households. Women were also sampled and interviewed, but from different 19 households than the men. Men interviewed men and women interviewed women in all locations except for Mexico, where the majority of the surveys with men were conducted by women. (See Levtov et al., forthcoming, and Barker et al. 2011 for more details on sampling and study design). The IMAGES study examines men’s attitudes and practices related to daily life, masculinity, employment, health, policies, fatherhood, sexual behaviors, and violence. While the survey was adapted slightly for each country, each questionnaire had approximately 250 items. The study protocol was approved by the ICRW institutional review board. 20 Figure 1 Details on data collection in each country (adapted from Barker et al. 2011 and Levtov forthcoming) Data Collection Bosnia Brazil Chile DRC Croatia India Mexico Rwanda Details Sample size, men 1532 750 1192 708 1453 1552 1002 2301 Site(s) Nationally One metropolitan Three Four areas: an One metropolitan Two metropolitan Three Nationally representative area: Rio de Janeiro, metropolitan internally area and two rural areas: metropolitan representative sample two neighborhoods: areas: displaced areas: Zagreb, & Delhi & Vijayawada areas: Monterrey, sample Maré & Vila Valparaíso, persons camp towns and villages in (Tamil Nadu) Queretaro, & Valquiere Concepción, & and a military two counties in Jalapa Santiago base, both Eastern Croatia in Goma and 2 rural villages south of Goma Sample Stratified by place of Two income groups: Stratified by Stratified by age Stratified by age and Census block Stratified by age Stratified by age stratification residence low income (Maré) and place of and place of place of residence selected by and place of and place of strategy middle class (Vila residence and residence (rural/urban) probability residence residence Valquiere), household socioeconomic proportional to size, (provinces) sample proportional to level systematic random size of community sampling to select household Female labor force 35% 60% 47% 70% 46% 29% 44% 86% participation 2008- 2012 Total Fertility 1.1 1.8 1.8 5.7 1.5 2.6 2.3 5.3 Rate, 2011 Ratio of females to 102 N/A 100 79 105 96 102 102 males in primary and secondary education (%), 2010 Specific legislation Yes Yes Yes No Yes Yes Yes Yes addressing Law on Protection from Maria da Penha Law Ley de Violencia Law on Protection The Protection of Ley General de Law on domestic violences3 Domestic Violence (2006) Intrafamiliar Against Domestic Women from Acceso de las Prevention and (2005) (No. 20.066, Violence (2003) Domestic Violence Mujeres a una Vida Punishment of 2005) Act, 2005 Libre de Violencia, Gender-Based 2012 Violence (No. 59, 2008) Percentage of female population age 15+ that supply labor goods and services production 2008-2012 http://data.worldbank.org/indicator/SL.TLF.CACT.FE.ZS 2 Rate for Bosnia and Herzogovena 3 World Bank, 2013) 21 Measures In this paper, we present descriptive, bivariate, and multivariate analyses examining the role of men and boys in preventing or promoting women’s agency and gender equality. See Appendix 1 for a complete description of variables used and variable construction. Below, we briefly describe variables used for analysis. Demographic variables We present and analyze various demographic variables, including age, education, marital/cohabitation status, number of children, and income. Given that actual age was not available for all countries, we used a three-category age variable: 18-28, 29-39, and 40-59. For education, the men were asked the highest grade they had completed and we categorized them as no formal education, some or completed primary, some or completed secondary school, or post- secondary. Because so few participants fell into the first category for some countries (i.e. only 0.1 percent of Bosnian men had “no formal education”), we collapsed the first two categories (no formal education and up to primary) for our multivariate analyses. We created a four-category income variable that captured relative income within each country: low-income quartile, mid-low income quartile, mid-high income quartile, high-income quartile (see Levtov forthcoming). This variable was intended to create income quartiles for each country where each category had roughly 25 percent of participants, but this was not always possible for countries like the DRC where income was asked as a categorical question. Marital/cohabitation status was measured dichotomously where men who were either living a partner or married to a partner were considered to be married/cohabitating. Number of children was reported as the number of biological children a man had. Outcome variables We conduct multivariate and bivariate analysis with two main dependent variables: 1) physical violence perpetration against women, and 2) active participation in prevention of violence against women. The violence perpetration variable is a dichotomous composite variable where if a man reports ever perpetrating any type of physical violence (slapping, pushing, hitting, kicking, choking, etc.) against a woman he is coded as a 1, and if he reports never having perpetrated any of those types of violence against a woman he is coded as a 0. The prevention of violence against women variable is based on a questionnaire item that asks, “What would you do if you saw violence being carried out by a stranger (man) against a woman?” Response options were 1) intervene during the episode, 2) speak to him after, 3) avoid/shun him, 4) call police, 5) do nothing, or 6) mobilize the neighbors. Because we wanted to identify men who would actively 22 prevent violence against women, men were coded as a 1 if they responded that they would intervene during the episode, call the police, or mobilize the neighbors. Other response categories were coded as a 0. Attitudinal and behavioral variables In an effort to investigate correlates of outcome variables across a wide range of countries, we were limited to only variables that were available in all (or almost all) eight countries and did not have a high proportion of missing data. Our main predictor variables were sexual violence perpetration, witness of intra-parental violence, GEM scale, attitudes towards violence against women, depression, binge drinking, fighting, and three variables on men’s awareness of violence against women campaigns and law. The sexual violence perpetration variable is a dichotomous variable created from nine items on different types of sexual violence. See Appendix 1 for full list of items. The Gender Equitable Men (GEM) scale was administered to participants in each country, but was comprised of slightly different items in each country depending on cultural relevance and factor analysis. The GEM scale measures the extent to which men agree with gender equality or separate roles for men and women, and has previously undergone psychometric testing for validity and reliability (Pulerwitz & Barker 2008; Shattuck et al. 2013). A higher score on the GEM scale indicates more supportive attitudes towards gender equity. In the multivariate analysis, we standardized the GEM score based on the mean and standard deviation for each country. The witnessing of intra-parental violence variable assessed whether or not a man saw his mother’s husband or boyfriend beat her at least once. Attitudes towards VAW were assessed by whether or not the man “strongly agreed” or “partially agreed” with the statement: “There are times when a woman deserves to be beaten.” Men who reported feeling depressed “often” or “sometimes” in the past month were coded as a 1 for the dichotomous depression variable. Binge drinking, also dichotomous, measured whether a man reported having 5 or more drinks on a single occasion at least once a month. The fighting variable measured whether or not the man had ever been involved in a fight with a knife or other weapon. The variables on men’s awareness measured whether or not men were aware of laws in his country on violence against women (aware1), aware of anti-violence campaigns (aware2), or aware of anti- violence advertisements (aware3). Additional attitudinal and behavioral variables are included in the descriptive analysis, see Appendix 1 for a list of how those items were coded. 23 Data analysis We present descriptive, bivariate, and multivariate analyses in this report. For descriptive tables, we used frequencies by country to describe participants’ demographics, attitudes and behaviors. We also examined bivariate relationships between physical violence perpetration and key demographic, attitudinal, and behavioral variables for each country. To assess whether or not these relationships were statistically significant, we used t-tests for continuous variables and χ2 tests for dichotomous variables. We conducted inferential analyses to examine the influence of selected variables on physical violence perpetration. Based on an examination of the literature and the variables available in the IMAGES dataset, we selected eight demographic, attitudinal, and behavioral variables as predictor variables in our logistic regression equation. First, using the model below, we ran a logistic regression model separately for each country. Then we conducted the analysis with all eight countries simultaneously using country fixed effects to examine the relationship between perpetration and the independent variables across all countries. ( ) where: θ = violence perpetration (dichotomous) x1 = age (categorical) x2 = education (categorical) x3 = income (categorical) x4 = employed (dichotomous) x5 = witness of intra-parental violence (dichotomous) x6 = attitudes towards VAW (dichotomous) x7 = standardized GEM score (continuous) x8 = fights (dichotomous) We then did the same with the dichotomous outcome variable of whether or not men report they would take action to prevent violence against women. Because this analysis was exploratory, we used theory and literature to identify independent variables to include in our model. Primarily, we were interested in what factors might be associated with men reporting a desire to prevent violence in their community. Like the previous regression analysis, we ran a logistic regression model separately for each country using the model below. Then we conducted the analysis with all eight countries simultaneously using country fixed effects to examine the relationship between taking action to prevent violence and the predictor variables across all countries. 24 ( ) where: θ = reporting the would take action to prevent VAW (dichotomous) x1 = age (categorical) x2 = education (categorical) x3 = income (categorical) x4 = employed (dichotomous) x5 = witness of intra-parental violence (dichotomous) x6 = attitudes towards VAW (dichotomous) x7 = standardized GEM score (continuous) x8 = fights (dichotomous) x9 = awareness of VAW laws (dichotomous) Observations with missing data on any of the included variables were excluded from the analyses (i.e., listwise deletion). Sample sizes included in the analysis are presented in the regression tables. Limitations The IMAGES dataset is a rich source of information on men’s attit udes and practices across the globe, but it does have its limitations. These data are cross-sectional so we are unable to make claims of causality with our analysis. Additionally, with the exception of Bosnia and Rwanda, the samples are not nationally representative so our findings are limited to the region or location where the data were collected. Because survey instruments varied slightly and were in each country’s language, there were certain variables that may have carried slightly different meanings in each location (despite double-back translation) limiting our ability to compare across countries. We were also sometimes unable to include variables in the multi-country analyses if they were not asked in every country. Previous studies have noted a concern about the accuracy of men’s self -reported violence perpetration (Archer 1999; Armstrong, Wernke, Medina, & Schafer 2002; Hilton, Harris, & Rice 2003; Yount & Li 2012). The IMAGES data collection procedures relied on men’s self-reports of violence perpetration. In Bosnia, Brazil, Croatia, India, and Mexico these questions were self- administered in private, but in Chile, DRC, and Rwanda the questions were administered by an interviewer. We are able to compare the percentage of men who report perpetration to the percentage of women who report victimization in the same region. Mexico, Bosnia, and the DRC are the only countries that demonstrate large discrepancies between women’s report and men’s report (see Table 9). In Mexico and Bosnia, the proportion of men reporting perpetration is lower 25 than the proportion of women reporting victimization. In the DRC the opposite is true: A higher proportion of men report perpetration than women report victimization. Though we do have concerns about the accuracy of men’s self-reported IPV perpetration from these countries, we believe the bivariate and multivariate analyses should still highlight important associations between perpetration and other variables in these countries. The other included variables are sometimes asked directly and rely on self-reported data and thus are subject to inaccuracies. Nonetheless, while some caution should be exercised in interpreting results given these limitations, the findings contribute to a thin evidence base on men’s roles in women’s agency and violence against women. RESULTS Basic demographic information for participating men from each country is provided in Error! Reference source not found.. All men were between the ages of 18 and 59. India had the youngest sample, with 47 percent of participants in the youngest category. Different age structure of respondents in each country could bias results since younger populations will have had less time to perpetrate violence and maybe be more progressive than the general population. Men’s educational achievement varied greatly by country. Men from the eastern European countries of Croatia and Bosnia were the most educated, with more than 60 percent attending secondary school and more than 30 percent attending some type of post-secondary education. India, Mexico, and Chile also had high proportions of respondents attending secondary school or beyond. Rwanda and Brazil comprised the least educated samples: 82 percent of the sample in Rwanda and 57 percent in Brazil had no schooling beyond primary. In the DRC, 42 percent had never attended school beyond primary. The percentage of men who were married or cohabiting in each country ranged from 42 percent in Chile to78 percent in Brazil. Men from the two African countries, Rwanda and the DRC, were the most likely to have children, with 84 percent of Rwandan men and 77 percent of Congolese men reporting having at least one biological child. Having six or more children was also much more common in those countries (Rwanda 20 percent, DRC 29 percent), whereas 2 percent or less had six or more children in the other six countries. In these latter countries, between 43 percent and 60 percent of men had no children. 26 Table 1. Demographic and other characteristics* Percentages exclude observations with missing data on that item. Bosnia Brazil Chile Croatia DRC India Mexico Rwanda n=1532 n=750 n=1192 n=1501 n=708 n=1534 n=1001 n=2301 Age n % n % n % n % n % n % n % n % 18-28 668 44 260 35 409 34 471 32 198 28 723 47 373 37 597 26 29-39 434 28 191 25 298 25 418 29 253 36 462 30 265 26 752 33 40-59 430 28 299 40 485 41 564 39 257 36 367 24 364 36 952 41 Education No Formal Education 14 1 21 3 17 1 1 0 84 12 180 12 14 1 406 18 1,48 Up to Primary School 87 5 401 54 127 11 55 4 213 30 133 9 116 12 64 1 Secondary School 1088 65 238 32 566 48 902 60 280 40 533 34 245 24 312 14 Post-Secondary 495 30 89 12 481 40 537 36 129 18 706 45 627 63 102 4 School Marital/Residential Status Married and/or 154 740 44 412 78 503 42 769 70 504 72 859 56 415 51 67 Cohabiting 5 Employment Status 123 219 Currently Employed 1083 64 563 75 850 72 1017 72 437 63 80 791 79 96 8 4 Number of Biological Children No children 1006 60 343 46 515 43 807 57 150 23 743 48 426 44 360 16 1 to 2 569 34 300 40 483 41 522 37 124 19 617 40 300 31 583 27 3 to 5 100 6 98 13 185 16 93 7 187 29 185 12 228 23 816 37 6 or more 9 1 9 1 8 1 2 0 188 29 7 0 17 2 434 20 Childhood Experiences Witness of intra- 155 11 108 16 354 32 235 16 285 43 579 38 161 17 876 44 parental violence 27 Income is difficult to compare across countries, as purchasing power and exchange rates can vary greatly. It should be noted that these income data are in quartiles for each country based on other respondents. Therefore, it is possible for example that those in the highest income quartile are actually in the lowest income quartile relative to the entire country population. For more details on the IMAGES respondents, see Barker et al. (2011), Dusanic (2012), Slegh et al. (2012), Levtov et al. (forthcoming). Attitudes towards gender equality The majority of men in all countries disagreed with the statement that “Rights for women mean that men lose out” (see Table 2). Less than 10 percent of men agreed with this statement in Chile, Croatia, and Mexico. Agreement with that statement was more common in India and the DRC with 44.2 percent and 38.2 percent agreeing, respectively. The majority of men in each country believed that “gender equality has already been achieved,” ranging from 57.1 percent of men in Mexico to 80.2 percent of men in India. In some countries, men’s attitudes towards gender equality differed somewhat by their age and level of education (See Appendix 2.). In all countries but India, education level was a significant predictor (p<.05) of men’s agreement with the statement, “Rights for women mean men lose out.” In each country, as education level increased, the proportion in agreement with the statement decreased. Differences in attitudes by age were significant only for Chile, India, Mexico, and Bosnia. In those countries, older men were more likely to agree than younger men. Table 2 Men’s attitudes towards gender equality, percent who agree or strongly agree Rights for Gender equality Gender equality women mean has come has been men lose out far enough achieved n % n % n % Bosnia 158 10.9 712 50.8 986 68.2 Brazil 78 10.4 353 51.2 470 68.6 Chile 99 8.3 754 68.6 665 60.3 Croatia 68 4.7 371 28.6 851 62.2 DRC 250 38.2 314 54.1 335 60.6 India 685 44.2 1205 87.0 1108 80.2 Mexico 70 7.1 717 74.5 538 57.1 Rwanda NA NA NA NA NA NA 28 Attitudes and practices regarding women’s roles and responsibilities While there was mostly broad support for gender equality, there was a wider range of views between countries on women’s roles and responsibilities (see Table 3). Only 9.9 percent of men in Brazil agreed that “changing diapers, giving baths, and feeding kids is a mother’s responsibility.” In contrast, 85.6 percent of men in India, 78.3 percent of men in the DRC, and 61.2 percent of men in Rwanda agreed that those tasks were a mother’s responsibility. In all countries except Croatia (35.8 percent), the majority of men believed that a “woman’s most important role is to cook and clean.” A minority of men believed that avoiding pregnancy was a woman’s responsibility; only in the DRC did a majority of men (60.9 percent) believe that it was a woman’s responsibility. In the other countries agreement ranged from 15.5 percent in Croatia to 49.4 percent in Rwanda. When comparing men’s responses to women’s responses on each of these items, we found similar patterns (see Table 6). Women in Rwanda and the DRC were the more supportive of these statements. A much lower proportion of Indian woman supported these statements compared to Indian men. Household decision-making is a key component of gender relations and women’s agency. When men were asked who had the final say on certain decisions within a household, men from most countries reported that either their partner/wife had the final say or that they shared final say equally with their wife/partner (See Table 4). Men from the two African countries (DRC and Rwanda) were notable exceptions. When asked who has the final say on “large investments such as buying a car, or a house, or a household appliance,” only 32.7 percent of Rwandan men and 39.8 percent of Congolese men reported their wife/partner having final say or sharing it with them. By contrast, the percentage in the other six countries ranged between 69.2 percent in India to 85.8 percent in Mexico. A similar pattern emerged when asking who has the final say regarding “food and clothing,” “the health of women at home,” and “the health of children in the home.” However, India was similar to the DRC and Rwanda for the decision-making related to the health of women, with a minority of men (40.5 percent) reporting their wife/partner playing a role in the final say. In general, women's reports of who has the final say are similar to men's reports. The one exception is Rwanda, where more women reported having the final say, or sharing the final say, than men reported. When looking at differences in men’s attitudes towards women’s roles and responsibilities by age and education (Appendix 2), education was a significant predictor in every country except for India and the DRC. For example, in Brazil, 65.8 percent of men in Brazil with a primary school education or none at all agreed that a woman’s most important role is to cook and clean. But only 22.7 percent of men with post- 29 secondary schooling agreed with that statement. The same trend was true in the other countries as well. Differences in attitudes by age were only significant in Chile and Bosnia, with older men reporting higher levels of agreement. Table 3 Men’s attitudes towards women’s roles and responsibilities, percent who agree or strongly agree A woman’s most Diapers/baths/feeding Avoiding pregnancy important role kids is mother’s is a woman’s is to cook/clean responsibility responsibility n % n % n % Bosnia 811 51.9 830 53.1 424 27.4 Brazil 401 53.6 74 9.9 270 36.2 Chile 647 54.4 543 45.6 553 46.5 Croatia 534 35.8 426 28.7 231 15.5 DRC 517 74.2 541 78.3 423 60.9 India 1373 88.5 1328 85.6 624 40.2 Mexico 557 55.6 257 25.7 220 22.0 Rwanda 1858 83.1 1367 61.2 1101 49.4 Table 4 Men who report that the final say belongs to their partner or jointly with their partner for four household decision-making areas: Final say on Final say on Final say Final say spending for spending for regarding regarding health food and clothing large investments health of women of children N % n % n % n % Bosnia 633 91.7 546 80.4 495 90.5 507 92.7 Brazil 327 76.9 325 76.5 386 91.5 232 86.3 Chile 571 88.3 520 79.5 439 83.5 462 86.4 Croatia 750 93.9 661 83.3 524 94.6 518 93.7 DRC 261 47.2 237 39.8 138 24.7 155 27.9 India 648 78.9 564 69.2 629 40.5 520 68.2 Mexico 510 92.9 465 85.8 412 88.8 431 92.9 Rwanda 805 38.5 680 32.7 637 32.6 875 45.4 30 Table 5 Women who report that the final say belongs to themselves or jointly with their partner for four household decision-making areas Final say on Final say on Final say Final say spending for spending for regarding regarding health food and clothing large investments health of women of children n % n % n % n % Bosnia 307 91.4 271 82.1 NA NA NA NA Brazil 190 76.0 187 74.8 235 94.0 299 91.2 Chile 240 96.0 213 85.5 236 94.0 261 88.2 Croatia 272 95.8 247 86.4 177 97.8 173 98.9 DRC 389 54.9 306 45.1 NA NA NA NA India 335 80.9 317 76.6 306 89.2 307 89.5 Mexico 279 90.9 252 83.7 284 92.5 294 95.8 Rwanda 715 59.2 607 50.9 633 52.8 691 58.7 Men’s attitudes regarding violence against women Violence against women is an important factor for limiting women’s agency, and men’s attitudes towards physical violence against women varied across the different countries (See Table 7 and Figure 3). Men from each country were asked whether they agreed or disagreed with the statement “there are times when a woman deserves to be beaten.” Overall, most men disagreed with this statement. In Mexico, only 5.8 percent of men agreed and in Chile only 10.0 percent agreed. Less than a quarter of all men agreed in Croatia (12.0 percent), Brazil (19.3 percent), Rwanda (20.5 percent), and Bosnia (23.1 percent). But, a majority of men agreed with that statement in India (64.8 percent) and the DRC (61.5 percent). Among women, Indian women (40.8 percent) and Congolese women (47.4 percent) were the most likely to agree with that statement. Similar to the men, a low proportion of women agreed in the rest of the countries (Table 6 and 31 Figure 3.). Attitudes towards violence against women are measured somewhat differently in the Demographic and Health Surveys (DHS), but for those countries with available data we found mostly similar results.3 Men from most countries also disagreed that a “woman should tolerate violence in order to keep her family together.” Fewer than 15 percent of men in the Latin American and Eastern European countries agreed, but more than half the men agreed with the statement in India (67.5 percent), the DRC (64.9 percent), and Rwanda (53.6 percent). For a breakdown of attitudes in each country by age and education groups, see Appendix 2. The extremely high share of Croatian women reporting agreement to the statement that “a woman should tolerate violence to keep a family together” is a notable outlier relative to other countries, and is much higher than for other types of regressive views held by men and women in Croatia. This merits further investigation to understand the measurement and cultural factors that might underlie this result. Men’s ideas about rape and the women who are raped can give insights into their conceptualization of what constitutes sexual violence against women (See Table 7.). Men were asked whether they agreed or disagreed with the statement, “If a woman doesn’t fight back, it’s not rape.” Chilean men were the least likely to agree (9.8 percent) and Indian men the most likely (66.0 percent). To better understand men’s opinions on this, we asked men whether or not they agreed that “In any rape case one would have to question whether the victim is promiscuous or has a bad reputation.” Men from the Latin American countries were the least likely to agree with the statement. Sixty-five percent of men in India agreed and nearly half agreed in the DRC (48.0 percent) and Bosnia (43.4 percent). Notably, awareness of VAW laws is not universal (Table 8). Croatian and Bosnian men were the least likely to respond “Yes” to the question, “Are there any laws in your country about violence against women?” (75.9 percent and 59.6 percent, respectively). About half or more of men in each country were aware of anti-VAW campaigns and anti-VAW media campaigns. 3 The DHS asks women a series of five questions about whether “A husband is justified hitting or beating his wife if” she a) “burns the food,” b) “argues with him,” c) “goes out without telling him,” d) “neglects the children,” or e) “refuses to have sex with him.” These were asked only of women, and only in the DRC, India, and Mexico. Percent who agreed for each question ranged between 28.2 percent and 55.5 percent in the DRC (2007 DHS), 14.1 percent and 34.7 percent in India (2006-06 DHS), and 18.8 percent and 43.6 percent in Rwanda (2010 DHS). 32 Table 6 Women’s attitudes towards physical violence against women and a woman’s role, percentage who agree or partially agree A woman’s There are A woman most Diapers/baths times when a should tolerate important role /feeding kids is woman deserves violence to is to mother’s to be beaten keep family together cook/clean responsibility n % n % n %t N % Bosnia 52 8.3 27 4.3 185 29.5 179 28.5 Brazil 41 9.2 26 5.8 167 37.3 116 26.0 Chile 32 7.6 31 7.3 213 50.1 199 47.0 Croatia 22 4.4 447 89.8 65 13.0 68 13.6 DRC 352 47.4 575 77.5 641 86.2 649 87.5 India 214 40.8 224 42.7 154 29.3 207 39.4 Mexico 12 3.2 9 2.36 224 58.8 119 31.2 Rwanda 255 20.0 870 67.9 1184 92.3 1157 90.2 Table 7 Men’s attitudes towards physical and sexual violence against women, percentage who agree (includes “partially agrees” for first two columns, “strongly agree” for last two columns) There are A woman If a woman You have to times when a should tolerate doesn’t fight question if a woman deserves violence to back, it’s not rape victim is to be beaten keep family together rape promiscuous n % n % n % n % Bosnia 360 23.1 204 13.3 364 27.8 552 43.4 Brazil 144 19.3 31 4.1 256 34.8 179 24.3 Chile 119 10.0 101 8.5 116 9.8 138 11.7 Croatia 177 12.0 86 5.8 234 17.6 360 27.3 DRC 424 61.5 451 64.9 299 45.6 293 48.0 India 1005 64.8 1048 67.5 1024 66.0 1009 65.0 Mexico 58 5.8 39 3.9 206 21.4 188 19.9 Rwanda 457 20.5 1198 53.6 NA NA NA NA Figure 2 Men's and women's attitudes towards a woman's roles and responsibilities by country 33 100 90 % who agree or partially agree 80 70 60 50 40 30 20 10 0 Bosnia Brazil Chile Croatia DRC India Mexico Rwanda A woman's most important role is to cook/clean: nMen nWomen Diapers/baths/feeding kids is a woman's responsibility: nMen nWomen Figure 3 Men's and women's attitudes towards VAW by country 100 % who agree or partially agree 90 80 70 60 50 40 30 20 10 0 Bosnia Brazil Chile Croatia DRC India Mexico Rwanda There are times when a woman deserves to be beaten: nMen nWomen A woman should tolerate violence to keep her family together: nMen nWomen Table 8 Men’s awareness of laws and campaigns related to violence against women, percentage who are aware Aware of anti-VAW Aware of anti-VAW Aware of VAW Laws campaigns or activities advertisements n % n % n % 34 Bosnia 1003 59.6 1125 73.6 1173 78.7 Brazil 679 90.7 131 17.6 308 41.3 Chile 909 78.2 578 49.0 993 84.0 Croatia 1110 75.9 1150 77.5 1260 85.0 DRC 542 82.3 508 78.8 520 79.5 India 1264 84.6 806 54.2 1165 78.2 Mexico 864 86.3 462 46.5 818 82.5 Rwanda 1896 82.4 NA NA NA NA Perpetration of violence against women Previous analyses of the IMAGES data (Barker et al. 2011; Levtov et al. forthcoming) have shown that men in the DRC (44.0 percent), Rwanda (38.7 percent), and India (37.4 percent) were the most likely to have ever perpetrated physical violence against a partner (Table 9 and Figure 4). Almost one third of Croatian men (32.6 percent) and Chilean men (29.5 percent) had perpetrated violence whereas a quarter of Brazilian men had (24.3 percent). Seventeen percent of Mexican men and 26.0 percent of Bosnian men reported perpetrating physical violence, though women in those countries reported higher levels. In all other countries, women’s reports of physical violence victimization were within 6 percent of men’s reports of perpetration. However, in Mexico and Bosnia, the discrepancy was 14 percent and 19 percent (in the DRC, fewer women reported experiencing violence). In Mexico, this may be due to the fact that many men in Mexico were interviewed by women, whereas in other countries only men interviewed other men. Women's reports of violence victimization in IMAGES is slightly different that reports from the DHS in the DRC and Rwanda, the only two countries with available data. In both cases, the DHS finds a higher percentage of women reporting violence victimization, likely due to the fact that DHS uses a difference sampling strategy, defines “ever partnered” differently, and uses more questions to identify violence.4 Table 9 Reports of men’s physical violence perpetration and women’s victimization Physical violence against a partner % of men who % of women who report ever report ever being 4The DHS uses seven specific questions to measure the DHS. IMAGES uses only five questions, combining two and leaving out one about “twisting her arm or pulling her hair.” Additionally, they ask women who have been married, rather than women who have ever had a partner, as IMAGES does. For more information see: http://www.measuredhs.com 35 perpetrating victimized Bosnia 26.0 44.6 Brazil 24.3 27.9 Chile 29.5 31.4 Croatia 32.6 38.1 DRC 44.0 21.9 India 37.4 30.7 Mexico 17.4 30.7 Rwanda 38.7 41.1 36 Figure 4 Percentage of men reporting lifetime perpetration of violence against a partner and women reporting lifetime victimization of partner violence by country 50 45 40 35 Men's (%) reporting 30 perpetration 25 20 Women's (%) 15 reporting victimization 10 5 0 We analyzed key demographic and behavioral variables to determine whether there was a statistical difference between men who had perpetrated violence against a female partner and those men who had not perpetrated violence against a female partner (see Table 10). We found that in all countries, the mean age of perpetrators was higher than non-perpetrators and this was statistically significant in all countries except for Brazil and Bosnia (data unavailable for DRC). We saw little difference between current employment status (employed or unemployed) and violence perpetration. Only in Mexico and Chile were these differences significant with perpetrators more likely to be employed then non-perpetrators. In some countries (Chile, Croatia, and DRC), perpetrators were significantly more likely than non-perpetrators to be married or cohabitating. In Mexico, the opposite was true: Non-perpetrators were more likely than perpetrators to be cohabiting or married (54.2 percent versus 43.2 percent). In all other countries there was no significant relationship between perpetration and marital status. We additionally examined whether there were differences between perpetrators and non-perpetrators in men’s attitudes and practices. When compared to perpetrators, men who were non-perpetrators were more likely to have a higher GEM score indicating greater support for gender equality (See Table 10 and Figure 5). This difference was demonstrated in all countries and was statistically significant in all countries except for India. Indian men had overall more support for inequitable norms, so the lack of significance may have been due to insufficient variation across the ranges of the GEM scale. While GEM score represents a man’s attitudes towards gender equality and gender roles, we also examined whether men’s 37 attitudes towards violence against women were significantly different between perpetrators and non- perpetrators. We found, as expected, that men who perpetrate violence are more likely to agree with the phrase, “There are times when a woman deserves to be beaten” (ATTVAW). This difference was statistically significant in every country. Figure 5 Gender Equitable Men (GEM) Score for perpetrators and non-perpetrators, by country5 50 45 Total Possible GEM 40 score Raw GEM Score 35 Non-perpetrators 30 25 Perpetrators 20 15 When looking at differences in men’s behaviors, we found some evidence that men who have perpetrated violence against a partner were more likely to have been involved in fights with weapons, more likely to binge drink, and more likely to be depressed. The differences in the proportion that have been involved in fights were significant in every country except for the DRC. Perpetrators were more likely to be depressed in all countries where data were available. The relationship between binge drinking and violence perpetration was only significant in Chile, Mexico, Bosnia, and the DRC. Finally, men who had perpetrated violence were more likely to report having witnessed their mother being beaten by a husband or boyfriend when they were children, and this was statistically significant in every country except for the DRC. Results of country-specific logistic regression analyses for perpetration of physical violence using 5 GEM scale is calculated differently for each country and therefore can only be compared within countries, not between countries (see Levtov et al. forthcoming for details) 38 variables with low levels of missing data are presented in Appendix 3. Results from all-country models with country fixed effects appear in Table 16. Using results from the overall eight-country sample, we found that age, income, witnessing intra-parental violence, attitudes towards VAW, GEM score, and being involved in fights were all significant predictors of having perpetrated violence against a partner (see Table 11 and Figure 7). The odds of ever perpetrating physical violence for men between 40-59 were nearly two times (OR: 1.88, 95 percent CI: 1.47 - 2.41) the odds of men between 18-28. Overall, income was not a significant factor, though individuals in the mid-high income quartile were slightly more likely to perpetrate violence than those men in the lowest income quartile. Men who held permissive attitudes towards VAW (i.e. “believed that there were times when a woman deserved to be beaten”) were nearly twice as likely to perpetrate VAW (OR: 1.70, 95 percent CI: 1.34 – 2.16). Men’s GEM scores were also significant predictors of violence perpetration. For every 1 standard deviation increase in men’s GEM score (indicating greater support for gender equality), men had more than 10 percent lower odds of perpetrating violence against a partner (OR: 0.89, 95 percent CI: 0.80 - 0.97). Men who witnessed their mother being beaten by a partner had more than 2.5 times the odds of ever having perpetrated violence against their own partners (OR: 2.53, 95 percent CI: 2.08 - 3.07). Finally, men who had been involved in at least one fight with a weapon had 2.38 times the odds of having perpetrated violence against women compared to those men who had not been in a fight with a weapon (OR: 2.38, 95 percent CI: 1.91 – 2.97). Notably, while we could not include depression as an indicator in the all-country model because it was not asked in two countries, it was a significant predictor of violence perpetration in all but one country (see Appendix 3). When controlling for all the other variables in the model, employment and education were not a significant predictors of perpetration of physical violence against a partner. 39 Table 10 Differences between men who have ever perpetrated violence against a partner and those who have not MEANS FREQUENCIES ATT AGE CHILD AGE GEM** VAW( DIFF5 DEPRESS BINGE FIGHT( EXP COHABIT EMPLOY %) (%) (%) (%) %) (%) (%) (%) Non-perpetrators 32.6 36.9 17.9 27.2 24.3 30.7 13.7 6.61 48.3 64.0 Bosnia Perpetrators 33.7 33.9 37.2 28.4 31.1 44.6 35.4 23.4 53.2 64.1 Non-perpetrators 36.2 26.3 16.0 36.1 6.1 67.4 14.5 12.5 79.5 74.6 Brazil Perpetrators 36.2 25.4 30.2 35.0 15.6 75.4 45.6 25.9 73.5 78.2 Non-perpetrators 35.3 38.0 7.2 18.7 14.5 35.7 11.0 23.8 40.4 69.9 Chile Perpetrators 39.1 36.4 16.3 27.1 21.1 45.8 26.2 50.0 51.5 79.5 Non-perpetrators 34.9 35.0 6.5 23.3 28.5 41.4 13.5 11.3 66.2 71.0 Croatia Perpetrators 40.5 33.4 22.9 28.3 38.8 44.9 27.6 25.6 75.4 73.3 Non-perpetrators NA* 25.4 56.2 NA NA 9.2 13.6 41.9 76.2 62.0 DRC Perpetrators NA* 24.1 71.4 NA NA 23.3 15.3 46.8 84.8 68.5 Non-perpetrators 35.8 23.0 61.3 50.7 22.3 21.2 3.6 30.1 88.3 95.4 India Perpetrators 38.7 22.4 70.3 56.8 32.9 26.7 12.4 63.7 89.3 96.0 Non-perpetrators 34.4 28.7 3.8 23.9 7.8 31.7 9.8 13.6 54.2 78.1 Mexico Perpetrators 36.4 26.5 15.2 24.4 17.4 45.0 24.4 37.0 43.2 86.6 Non-perpetrators 39.7 27.6 15.6 NA NA NA 3.8 36.6 90.3 97.7 Rwanda Perpetrators 40.8 26.9 23.4 NA NA NA 6.5 57.7 89.0 98.0 Bolded means significant differences between perpetrators and non-perpetrators, p<.05 *In the DRC, age was collected only in intervals, see demographic Table 1 **GEM is calculated differently for each country and therefore can only be compared within countries, not between countries (see Levtov et al. forthcoming for details). ATTVAW=agree or strongly agree that there are times when a woman deserve to be beaten, AGEDIFF5=for men who have a current partner, the man is at least 5 years older than the woman, DEPRESS=depression variable, BINGE=at least one binge drinking episode in last month, FIGHTS=has fought with weapons, CHILDEXP=exposure to family violence, COHABIT=marital/cohabitation status, EMPLOY=whether or not employed 40 Figure 6 Intersection of violence perpetration, Inequitable GEM score, and Permissive attitudes towards VAW (only for men with responses on each item), all countries We found men's GEM score and attitudes towards VAW to be significant predictors of violence perpetration. Below, in the Venn diagram, we demonstrate that there is substantial overlap between men with an inequitable GEM score (defined as 1 standard deviation below or lower on the standardized GEM measure in each country) and permissive attitudes towards VAW. Additionally, 34.5 percent of men who perpetrated violence reported having permissive attitudes towards VAW and the rest (65.5 percent) of perpetrators did not report permissive attitudes towards VAW. Looking at it from another angle, more than half of men with permissive attitudes towards VAW have never perpetrated violence. While the relationships are strong in the multivariate model, this diagram indicates the substantial complexity in determining violence perpetration. Men without inequitable GEM score, permissive attitudes towards, Men with inequitable or reported violence GEM score perpetration (total: 17%) (total: 49%) 5% 3% 5% 4% 18% 8% 5% Men who have ever Men with permissive Perpetrated Violence attitudes towards VAW (total: 31%) (total: 24%) Total N = 8,702 41 Table 11 Correlates of physical violence perpetration against a partner, presented as adjusted odds ratiosa Adjusted 95% Demographic and ORa Confidence Predictor Variables (n=7810) Intervals Age 18-28 (REF) 1.00 -- Age 29-39 1.56*** 1.35 - 1.80 Age 40-59 1.88*** 1.47 - 2.41 No schooling or primary (REF) 1.00 -- Secondary school 0.95 0.75 - 1.21 Post-secondary school 0.76+ 0.55 – 1.04 Low income (REF) 1.00 -- Mid-low income 1.11 0.93 - 1.33 Mid-high income 1.17** 1.04 - 1.31 Highest income 0.96 0.75 - 1.22 Employed 1.08 0.94 - 1.23 Witness of intra-parental violence 2.53*** 2.08 - 3.07 Permissive attitudes towards 1.70*** 1.34 – 2.16 VAW GEM Score (standardized) 0.89* 0.80 - 0.97 Has been involved in fights 2.38*** 1.91 – 2.97 + p<0.10, * p<0.05, ** p<0.01, *** p<0.001 (Bolded means significant P<.05) a Adjusted for all other variables presented in table 42 Figure 7 Visual representation of significant correlates of violence perpetration by country ALL Bosnia Brazil Chile Croatia DRC India Mexico Rwanda Country Age Education Relative Income Employment Witness of Intra- parental Violence ATT VAW GEM score Depressed NA NA NA Involved in Fights Non-significant (p>.10) Almost significant (0.05